Ron Vale (UCSF, HHMI) 2: Molecular Motor Proteins: The Mechanism of Dynein Motility

Hello. I'm Ron Vale from UCSF. In part one of my iBiology talk, I introduced biological motility and I focused on the mechanisms of kinesin and myosin, and in this talk I'd like to discuss our more recent research on the mechanism of dynein motility.

Now, this year, 2015, is the first anniversary of the discovery of dynein, which was made by Ian Gibbons, and Ian described a new type of ATPase from cilia that was involved in powering its motility. Now, even though dynein was discovered twenty years before kinesin, we know a lot more about kinesin motility than we do about how dynein works. And the reason for that is that dynein is an incredibly complicated machine.

First of all, it's a massive protein complex, one of the largest in the cell. It's about one and a half megaDaltons in size. Even the gene that encodes the motor polypeptide is very large.

The motor polypeptide is about 500 kiloDaltons in size, so it's one of the largest polypeptides in the genome. Even the motor domain itself is massive, shown here. It's about eight times larger than the kinesin motor domain.

So simply because this motor has been so big and complicated, it has made it difficult to study. It's been difficult to express it and obtain pure protein, difficult to manipulate it by recombinant protein techniques, and also non-trivial to get a structure by X-ray crystallography. So all of these have posed challenges.

And in 2007 I actually gave an iBiology talk, which you can find in the archives, and at that moment in time we were primarily studying the single molecule motility of yeast dynein, but we knew very little about the structure of dynein in any great detail. Well, a lot has changed in those intervening years and I'd like to share to share with you today recent progress that's been made on the dynein structure, and now we do have atomic structures for dynein and we're beginning to get some insight into how this motor works. So first of all, let me just tell you about the different kinds of dyneins.

A major class of dynein are the axonemal dyneins. These are dyneins that power the movement of cilia, they also power the movement of flagella from a sperm, for example, and this just shows what the architecture of the axoneme looks like. It's composed of nine unusual types of microtubules that are called outer doublet microtubules, and they have this particular structure here.

And on these outer doublet microtubules sits the dynein molecule. In fact there are two different kinds of dyneins -- there's an outer arm dynein and an inner arm dynein -- and they sit statically on one of the outer doublets, the A tubule, and then they reach across to the neighboring B tubule and they cause a relative sliding between these adjacent microtubules. Now, this sliding motion of the two microtubules in the cilia gets converted into this sinusoidal beating pattern by a process that's still very poorly understood.

Now, in addition to these axonemal dyneins, there are cytoplasmic dyneins that are more cargo-transporting motors. So, one of these is cytoplasmic dynein 2 and it's responsible for a particular type of cargo transport that actually occurs inside of cilia and flagella, and it's responsible for transporting, first of all, building blocks up and down for the maintenance of cilia and flagella, and also some kinds of signaling molecules are also present in cilia, and they are trafficked by motor proteins as well. This shows an image of what cargo transport looks like, what intraflagellar transport looks like, by fluorescently tagging one of these cargo subunits.

Now, a kinesin motor transports the cargo up to the tip, to the plus end, but dyneins are minus end-directed motors, and so the dynein transports cargo in the opposite direction, from the tip towards the cell body. Now, there's also a cytoplasmic dynein 1, and that's responsible for virtually all of the cargo transport that occurs in the cytoplasm of cells towards the minus end. So kinesins, as you remember from the first lecture, move things out to the plus end, and this one type of cytoplasmic dynein carries out virtually all of the transport in the opposite direction, towards the minus end.

So, things that are transported include membranes, mRNAs, protein complexes are transported, as are viruses. Here's just one example, here, which is a melanocyte cell. This is a skin cell that carries pigments which are melanosomes, and in some organisms like amphibians and fish they can change the distribution of their melanosomes, so that when they're distributed outward the skin color is dark, when they're moved inward the skin color turns lighter, and this transport towards the center that you see here is driven by cytoplasmic dynein.

Now, in the first iBiology talk, I told you that myosin and kinesin, in fact, are relatively similar to one another. They're similar in structure and they clearly evolved, some time in evolution, from a common ancestor. But even though kinesin and dynein both operate on microtubules, they're not at all related to one another.

In fact, dyneins emerged from a completely different evolutionary lineage of ATPases and they belong to a group of ATPases that are called the AAA ATPases. And in fact dynein is a rather unusual member of this AAA ATPase family. Most of them are not traditional motors that you think of in terms of moving along a track.

Instead, they use ATP energy to produce mechanical work on molecules like proteins to basically break them apart and unfold them. So, an example that occurs in bacterial and eukaryotic proteolysis is that there's AAA ATPases that sit on top of the proteolytic chamber and their job is to take an incoming polypeptide and basically unravel it and stuff it through this hole into the proteolytic chamber so that the polypeptide can be degraded. So, let me tell you a few things are kind of more universal about these AAA ATPases.

First of all, most AAA ATPases encode a relatively small protein that has two domains, a large domain and a small domain. This is the basic ATP binding unit, but this single subunit is not the functional element of how these proteins work. They self-assemble, oligomerize into a hexamer, and it's this hexamer that's the active agent.

And in fact adjacent subunits help one another to hydrolyze the ATP, and in the last example thing ring-like structure is what actually unfolds the polypeptide and stuffs the polypeptide into this chamber that you see here. Now, dynein again is unusual in the fact that it makes a ring of AAA ATPases, but it does so by placing all the AAA domains in one very, very long polypeptide chain. And this just shows the motor domain structure of dynein and shows the positions of the six AAA domains.

And because they're in one polypeptide, they've each evolved different amino acid sequences over time and have evolved different functions. So AAA1, for example, is the main ATPase site of dynein, so this is what is really responsible for driving motility, as I'll show you later. AAA2 binds ATP, but it doesn't seem to be.

. . bind it in a cyclic or hydrolytic manner.

AAA3 also plays an important role that I'll come back to later and it may be a mechanism of regulating dynein. AAA4 also hydrolyzes [ATP], but it seems to play a very minor role in dynein motility and one that we don't completely understand. So, I'd like to address this subject now of how dynein can move along a microtubule, and in addressing this problem one has to tackle it using different kinds of techniques that are complementary.

So, one approach is to measure the motility of dynein, particularly at the single molecule level, and this gives you all kinds of information about the dynamics of dynein and how it's stepping along the microtubule track. But it's relatively low resolution information, in other words it can't really see the protein structure and what it's doing. On the other hand, we can do X-ray crystallography or do electron microscopy and these give higher resolution information, even down to atomic detail, but they're static images, so, you know, we see the protein frozen in time and it doesn't provide the information on the dynamics.

So, somehow to piece together the answer to this problem, one has to use the information from both of these techniques and try to work out a model of how dynein might work. So, let me tell you first about in vitro motility assays. This shows an in vitro motility assay for yeast cytoplasmic dynein, where we've labeled the dynein with a fluorophore, and you can see these individual dynein molecules moving beautifully along these microtubule tracks.

It's processive movement, meaning the dynein can take many steps along the microtubule track without letting go. Now, we can also measure this motility with greater precision if we use a computational approach. Basically, each of these individual spots, fluorescent spots of dynein that you see.

. . as they pass through the microscope, the light spreads out to what's know as a point spread function, so they appear.

. . these single fluorophores appear to have a diameter of about 250 nanometers.

But if you collect enough photons, you can describe that fluorescence, that spread out fluorescence intensity profile, and you can fit that intensity profile with a Gaussian curve, and the center of that Gaussian curve defines kind of the midpoint of where that fluorescent spot is. Now, you can take successive snapshots of dynein moving along the microtubule and at each snapshot you can mark the position of that centroid, and that's what all these individual dots are here, data points are, and this is for a kinesin molecule, but you can see for example, here, the motor protein was pretty much stationary on the track and then it took a jump forward, so it took an abrupt step forward along the microtubule track, and this kind of mechanism allows you to get a great deal of information on the stepping behavior of the motor on the microtubule. And I should say that this general method was first developed by Ahmet Yildiz and Paul Selvin.

So, let me first describe to you how kinesin steps along the track for comparison with dynein. So, kinesin always walks in this hand-over-hand manner, where the front motor domain. .

. these two motor domains are identical. .

. but the front one undergoes a conformational change and that causes the displacement of the partner head from a rear site to a forward site. And this is how kinesin walks for long distances in this kind of very regular hand-over-hand manner where it's stepping from one tubulin subunit to the next.

And you can even see this if we label the two heads with two different fluorescent dyes. So, we marked the two heads by a red color and a blue color and now we plot the position of these heads as they're stepping along, and you can see here, for example, in this frame over here, the red head is in front of the blue head, just like this diagram, but then the blue head leapfrogs past the red head and now the red head leapfrogs past the blue head, etc, etc, and you can see how these two heads are exchanging position in a regular, alternating manner. Now, dynein stepping doesn't look anything like this, so a similar experiment of marking the two dynein heads with two different dye colors was done by Ahmet Yildiz and Sam Reck-Peterson.

Both Ahmet and Sam were postdocs in the lab, but the work that I'm showing you here was done in their independent laboratories at Berkeley and Harvard. So, what you can see here is if we look at the position of the blue head, here, in step number 1, it's taken a big step forward, but in step number 2, instead of the partner taking the step, that same head now has taken yet another step along the microtubule. That is step number 2.

And now, finally, the rear head, in step number 3, begins to catch up, but it doesn't pass the blue head. And now in step number 4 the blue head still takes another step forward. So, what you can see from this is that the dynein is exhibiting an inchworm pattern, where the two heads can maintain their front and rear position and both step forward together.

And second of all the two heads are not necessarily exchanging roles in timing of stepping. Here, for example, the blue head took two successive steps before the red head took a step. So, this is just a very different kind of motility, an irregular motility that's not present in kinesin.

Also, dynein can take very different sized steps as well, so for example, here, here's a very large step of dynein going forward, but these steps here are smaller, so the step size of dynein is not as regular as kinesin. Furthermore, if you look at this trace, there are many times when dynein is actually taking a step backward before it takes a step forward, and these backward steps are fairly frequent for dynein and very rarely seen for kinesin, especially if kinesin is not trying to work against the load. So, let me just review the things that I just told you.

Kinesin has a very regular step size, this is the distance between subunits on the microtubule track, dynein more variable. Kinesin has this hand-over-hand stepping. Dynein can exhibit this as well, but it also exhibits this inchworm pattern.

The two heads of kinesin take turns moving; that is not necessarily true with dynein. And while backwards steps are rare for kinesin, as I showed you they're quite frequent for the dynein molecule. So, now I'd like to go on and discuss: How is it that dynein can actually take these steps along the microtubule track?

What is the structural basis for this movement? Well, a first big breakthrough in this problem came from pioneering electron microscopy studies by Stan Burgess, and this shows the images that they got of dynein in two different nucleotide states, and from these EMs (electron micrographs) you can see, for example, the ring of these AAA ATPase domains, but you can also see a couple appendages. One is a long stalk that comes out of dynein that leads at the very tip to its microtubule binding domain, and there's another appendage that you can see here as well.

This is something that they termed the linker. It's something that sits kind of across the ring and then extends out of the ring. And what they noticed in these two different nucleotide conformations is that the position of the linker relative to the ring and to the stalk can change.

So, here it's sitting. . .

it's emerging from the ring far from the stalk and here they're merging close together. And they thought that this motion of the linker may act kind of like a lever arm or a mechanical element similar to the lever arm of myosin, so what they proposed is that the motion of the linker relative to the ring might be able to generate a force upon a microtubule that would cause it to slide, and I'll come back to this later. So, of course we had to get higher resolution information of dynein and that had to be derived from X-ray crystallography, and it was quite a struggle to get a crystal structure of dynein and in fact our lab was able to get the first crystal structure of dynein in a nucleotide-free state in 2011, but shortly thereafter a whole bunch of other nucleotide conformations of dynein were reported.

So, the group of Kon and colleagues from Japan reported a very nice structure of Dictostelium cytoplasmic dynein with ADP, and in the last year or two our lab got a structure of dynein with an ATP analogue called AMPPNP, and Andrew Carter's lab was able to get a structure with ADP-vanadate, which may be mimicking an ADP-Pi state. And what we'd like to do is kind of similar to what you see in this image of the horse here, where you could see different snapshots of the horse taken as it's executing a gallop and from these different snapshots you can see the different conformations of the horse and begin to piece together how this horse is able to execute motility, and by the same principle we're trying to use these snapshots of dynein to understand how it changes its conformation in order to execute motion. So, now I'd like to give you kind of a tour of what we learned about the crystal structures, not just from our lab but from all the crystal structures that have emerged from the field.

First of all, here's just an image of dynein compared to kinesin, and you can see how much bigger dynein is compared to kinesin and how much more complicated a motor domain it is. And here's the position of the different AAA domains that I showed you before in this linear diagram, but here's how they map out on the dynein motor protein, and they're all color coded in the same way that you see in this linear diagram, here. So, I'll focus on a few important components.

. . so, the first is AAA1.

So, this is, again, the main hydrolytic site. If you make a mutation in AAA1, you completely knock out dynein motility, and interestingly this AAA1 is actually the region that's farthest away from the microtubule. Now, the other domain that I.

. . AAA subunit that I mentioned that's important is AAA3, and this is its position over here.

As I said before, it also hydrolyzes ATP and plays an important role in the mechanism, and I'll explain how it works later in this talk, but if you prevent ATP hydrolysis by AAA3, dynein isn't completely inactive, but the velocity of movement goes way down with a hydrolysis mutant. So, here's now an atomic resolution image of the linker that I described before as a mechanical element, and here it's shown extending across the ring. Here is the microtubule binding domain that's a small domain that interacts with the microtubule, and in between the ring and the microtubule binding domain lie these two coiled-coils.

One is called the stalk, but there's a second coiled-coil called the buttress, which in fact extends out of the ring and makes an important interaction with the stalk that I'll describe in a second. So, one of the interesting things that we want to know from this structure is how information, or conformational changes, are propagated to control various aspects of dynein function, and this is a particularly fascinating question for dynein because we know that when ATP binds at the very top of this molecule over here it has to relay a conformational change all the way down to the microtubule binding domain, which in fact causes this microtubule binding domain to release from the microtubule so it can step forward along the track. So, how this propagation occurs is a fascinating question, especially over this long distance of about 25 nanometers.

We also know that the ATP binding must be transmitted also to somehow change the conformation of where this end of the linker is going to be positioned on the ring. So, I'd like to now share with you some ideas of how we think this long range conformational change works, based upon this collection of new X-ray structures that were obtained. So first of all, let me just tell you a hint that we had from our first X-ray crystal structure in the nucleotide-free state, and this just shows the AAA ring, just focusing on the large domains.

And the one thing that you notice here is that this ring is not symmetric, it's a very asymmetric structure and there are a couple gaps in this ring where the AAA domains are farther apart. And this gap between AAA1 and AAA2 was particularly interesting and also surprising, because this is the region where ATP binds and drives motility, but we know from other AAA proteins, ATPases, that for ATP to be hydrolyzed, these two domains, AAA1 and AAA2, have to come closer together because there are residues that contribute to the hydrolysis both from AAA2 and AAA1. So we speculated, although we just had one nucleotide state here, that what may happen in dynein motility is that in the nucleotide free state there's a large gap, but when ATP binds that gap closes, and that closure then propagates a conformational change around the ring that gets transmitted to the microtubule binding domain and also gets propagated to the linker to change the linker conformational, all, though, initiated by the binding of ATP and the closure of this gap.

So, I'll show you that these general ideas appear to be true, and what you're seeing here is a morph, so we're going to slowly go from one crystal structure, which is the ADP structure from the Kon et al lab to another crystal structure, which is ADP-vanadate, which is more like an ATP state. So, this is the conformational change that presumably happens with ADP is exchanged for ATP in AAA1. So, when ATP binds to AAA1, you'll see a conformational change and in this video I'm going to focus particularly on these coiled-coils, and how the conformational change can be propagated from AAA1 all the way down to the microtubule.

So, here's the movie. You can see the whole ring kind of distorting in shape, and if you look at what happens here, this orange coiled-coil, the buttress, gets pulled away from the stalk, so that creates tension on the stalk, over here, and that does something interesting to the two helices that make up the stalk. It causes a sliding motion to occur so that the two helices can move a short distance relative to one another, but that sliding motion gets propagated all the way down the coiled-coil, all the way to the microtubule binding domain, and causes a subtle change in the microtubule binding domain structure that changes its affinity for microtubules.

And in fact this kind of mechanism was speculated many years ago by Ian. . .

in 2005 by Ian Gibbons and colleagues, and now it looks like there's good structural evidence for this as well as other types of evidence that has been obtained by other laboratories, including Kon and Sutoh. So, I now want to also focus this same morph between these two nucleotide states, but with reference to the linker. And you'll see that when ATP binds to AAA1, you'll see the change in the AAA subunits, and now we'll focus on what's happening in this linker, and you can see it undergoes this large conformational change, effectively going from a straight state to this bent conformation.

So, earlier in this talk, I described single molecule motility studies that provide information on how the dynein motor steps along the microtubule track and then I described X-ray crystallography and EM studies that provide information on conformational changes that occur in the dynein motor domain, and now what I'd like to do is to synthesize both pieces of information together into a model that describes how dynein is able to move along a microtubule. And this model is presented in the form of an animal that's made by Graham Johnson. Many parts of this animated model are speculative at the present time and no doubt, as we get more information on dynein, this model will change over the years.

But for right now it's useful as a way of synthesizing data that's been gathered by many different laboratories on dynein, and also to generate models for dynein motility that can be tested in the future by experimentation. So first of all, let me show you what you're going to see in this movie. This image that you see here of the dynein dimer is derived from X-ray crystallographic data.

However, we don't know very much about how the two dynein motor domains are connected to one another or how they're attached onto a membrane cargo, for example. So this part of the dynein molecule is more stylistic and simple because we simply don't have that structural information right now. Now, when I start playing this movie, you can see the dynein jiggling back and forth.

This jiggling is due to Brownian motion, which is driven. . .

thermally driven collisions of water molecules with the dynein, in fact this Brownian motion is probably much more vigorous than shown here in this animation. Here are the different parts of the dynein. Here's the ATPase ring, the stalk that connects the ring to the microtubule binding domain.

Here, colored in dark blue, this is the strong binding state of dynein. You'll see it transition to a light blue color when it undergoes a transition to a weak binding conformation. You'll see conformational changes occurring in the linker that I already described and that transition will be shown from a change in color from this yellow state to a red state.

And when the microtubule binding domain is detached you'll see it also jiggling, kind of moving randomly back and forth along the microtubule. That again is due to Brownian motion and it probably helps this microtubule binding domain execute a search for new binding sites along the microtubule lattice. So now, let's start this movie and watch how dynein steps along the microtubule.

And I'll show you the first step and then we'll analyze it in greater detail in the second step. So, here this leading head takes a step forward, it's jiggling around and now it redocks onto a microtubule binding site. Now you'll see the rear head take a step.

It took a step forward and you can see this linker undergo a conformational change from this yellow state to this red state, and this conformational change is accompanied, we think, potentially, by a rotation of the ring, and this rotation of the ring can change the angle of the stalk, pointing it and the microtubule binding domain forward on the microtubule track, which then allows this microtubule binding domain to reattach to a tubulin subunit farther towards the minus end of the microtubule. So that's what you'll see in this next step. It's going to redock, right there, and once it rebinds that is accompanied by, we believe, hydrolysis of ATP and the release of phosphate from AAA1, and that release of phosphate causes this conformational change, again, of the linker from this bent red state to this straighter yellow state, and this conformational change, we also think, may produce a tug on the cargo that advances the cargo forward along the track.

So, now let's see these conformational changes again, in this next sequence, and you'll also see the different types of dynein stepping in this next part of the movie. So here, the leading head steps forward, again, takes a big step forward. It redocks, but now it actually takes a step backward along the microtubule track.

Here's the rear head, it actually, by Brownian motion, scoots around the other head in this hand-over-hand motion. It now takes another step forward along the microtubule track, and now its partner head again undergoes a conformational change and takes a step forward along the track. And we think by this kind of process the dynein molecule is able to progressively move along on the track, and now let's have another look at this video and see all these steps in action one more time.

Now let me come at the end to this other AAA domain, AAA3, and let me tell you how we think that works. As I said, this also plays an important role in motility, and in particular we know if we block ATP hydrolysis, the motor stops working. And the mystery was why that was true, because we know that hydrolysis in AAA1 is sufficient to do all the conformational changes of the linker and for dynein to take a step forward, so the reason why AAA3 seemed to be important wasn't that clear.

But an answer to this came from structural studies from our lab, Gira Bhabha and Hui-Chun Cheng, where they looked at the conformation and the conformational changes of dynein, not so much when there are different nucleotides in AAA1, but in two different nucleotide states in AAA3. So, in particular, comparing when ADP in bound in AAA3 versus when ATP is bound in AAA3. And I'll show you, when AAA3 is in these different nucleotide states, what happens to the conformational change that occurs when ATP binds to AAA1.

So, the first is the movie you just saw. That is the conformational change that I showed you where the linker undergoes this large conformational change and the whole ring changes its structure. But, if we now load AAA1 with ATP, but now there's ATP in AAA3 as well, what you'll see is a very different picture.

The conformation of this side of the ring changes, you can see a dramatic conformational change there, but the conformational change stops at about AAA4 and doesn't get propagated around the rest of the ring, and never causes a conformational change in the linker or in the stalk domain. So AAA3 is in effect blocking the conformational change and preventing it from propagating throughout the ring. So the way I like to think about this is that AAA3 seems to be like a gate that controls the propagation of conformational change throughout the dynein ring.

AAA1 is the trigger, so in this image of dominos here, it's what initially kicks off the chain reaction that moves from one AAA domain to the next, and eventually can move all the way down through the ring to AAA6 and cause this massive conformational change. But if AAA3 has ATP in this site, it actually acts to block the propagation. It's almost as if I have a finger holding this domino down and preventing the propagation from going any further.

So the blocking of the conformational change, or the release to allow it, seems to be the primary activity of AAA3. So, that gives an update of what we've learned about dynein in the last few years, but I must say we're still very much at the beginning and there are a tremendous number of unknown questions. So, I illustrated some atomic structures and conformational changes that occur, but we don't really know how those structural changes relate to the stepping of dynein on the microtubule.

What would be particularly nice is instead of just getting static images of dynein, we can actually monitor and measure dynein structural changes while it's in the act of motility. And there are ways of doing this, for example techniques such as single molecule FRET, which act as probes to measure certain conformational changes that occur in a protein, and perhaps those kind of techniques can be applied to dynein so we can actually see dynein stepping and simultaneously measure conformational changes. I also gave you structural information on the role of AAA3, showing that it can block a conformational change of dynein and thereby prevent its motility.

But we don't really understand how and why AAA3 does this. How does the cell use this control mechanism to regulate dynein motility? How does it actually control whether AAA3 has an ATP or an ADP in the active site?

So, we have no idea on this issue right now, and this is obviously going to be important for understanding what the real purpose of AAA3 is in dynein cell biology. So, with that, I'd like to thank the many people that contributed to this work. First of all, people that were in the lab previously, a fantastic group of individuals that helped launch the dynein project in the lab -- Sam, Ahmet, Andrew, and Arne -- now have all gone off to their own labs and are very successful, and I've discussed a lot of their work from their independent labs in this talk.

And Carol Cho was a graduate student who has now gone on to Korea. And the more recent work is the work of Gira Bhabha and Hui-Chun Cheng. Gira is still in the lab and Hui-Chun has moved to her own lab in Taiwan.

And with that I'd like to thank you for your attention, and in my third iBiology talk I will discuss the regulation of mammalian cytoplasmic dynein.